The AI suite of City Research Solutions · human research since 1979 →

Synthetic research that answers like your customers, because it was built from them.

Calibrated to your fielded studies, error published.

Most synthetic-consumer tools generate respondents from a demographic profile, with no evidence that they answer the way your actual customers do. PERVASIVE INSIGHTS™ is built the other way round: a synthetic panel calibrated to how YOUR customers actually answered in 40+ years of human research, with the calibration error published topic by topic — plus every study you have ever run, in one searchable corpus. Interview it, survey it, or sit it around a table, in minutes.

Survey results · Compare to human
Q1: Which of these fresh dog-food packages would you pick up first? 200 synthetic respondents · shown the four packages below
Canine ChoiceHappy PawsNature's HarvestCascade Provisions
Cut by:OverallBy segment▣ Compare to human
Closest comparable human question
Which of these dog-food packages would you pick up first? (shown four shelf packages)
Package appeal · Fresh Pet Food Concept Test 2025 · n=820 · q14 · single_select_pct
PackageSyntheticHumanΔ
Canine Choice34% n=6832.1%+1.9
Happy Paws29% n=5827.6%+1.4
Nature's Harvest21% n=4223.9%-2.9
Cascade Provisions16% n=3216.4%-0.4
within 5 ptsbeyond 5 ptsAvg Δ 1.6 pts · Calibration state: Measured

Your own human data, layered into every synthetic result. Illustrative figures, sample conditions.

Synthetic research is only as good as the real research underneath it.

There are three kinds of "synthetic." Generated personas are built from demographic filters alone: no customer data, and no evidence of how closely they track real respondents. Data-described personas are built from web analytics or CRM records: they know who your customers are, but their answers are still guesses. Calibrated synthetic panelists are built from your fielded research and checked against it, topic by topic, segment by segment, on a report card you can open any time. Pervasive Insights™ is the third kind. Machine learning is the method; your research is the material.

1
Generated personasBuilt from demographic filters alone. No customer data, and no evidence of how closely they track real respondents.
2
Data-described personasBuilt from web analytics or CRM records. They know who your customers are; their answers are still guesses.
3
Calibrated synthetic panelistsBuilt from your own fielded research and checked against it, with the error reported. Where you have asked the question before, the human answer sits beside the synthetic one.

Two halves, five ways to use them. One half holds every study you have ever run — decks and full datasets — in one indexed corpus you can question in plain language. The other is the calibrated panel of your buyers, ready for any question you would put to a real customer. Every output is cited or calibrated back to the data it came from.

What is actually new: you have been able to analyze your own data for a decade. What you could not do twelve months ago is hold every study you choose to put in — decks and full datasets — in one indexed corpus, with an AI agent over the top, and ask one question across all of it.

Part one

Your own research

— everything you've already learned, made instantly usable
The Library
Research Library

1 · Library search

Every study you've ever run, indexed in one place and searchable in plain English. One question is answered from across all of them — with a citation back to the exact slide it came from.

Try asking: "What have past studies said about why buyers pick one brand over another?"
The Data
Ask Your Data

2 · Data synthesis

Work straight from your raw datasets — all of them at once. One question runs across every matching dataset in your corpus, three waves of a tracker in a single pass, and comes back as one finding with the chart. Then narrow it: by segment, by age, by wave.

Try asking: "Which has cost us more awareness since the peak — buyers under 40, or over 40? Now cut it by income."
Part two

Your synthetic panel

— your buyers, calibrated, ready for any question
Go deep
One-on-One conversations

3 · One-on-One

Talk with a single calibrated customer profile. One person, one conversation, in their voice — for pressure-testing language or a raw gut reaction.

Try asking: "Why did you choose that model over the competitor's?"
Go wide
Synthetic surveys

4 · Surveys

Field a full survey to 50–500 calibrated respondents and get structured, quantified results back in minutes instead of weeks — fast enough to test several ideas at once.

Try asking: "Run aided awareness and rank these four concepts across every segment."
Hear the room
Synthetic focus groups

5 · Focus Groups

Sit in on a moderated round-table of your customers. Watch opinions form and shift, catch disagreement, pull quotable verbatims — as a transcript or a report.

Try asking: "Run a group on the new concept and surface the main objections."

Five tools. One real sample of each.

This is what each tool actually returns — the same kind of deliverable your team already knows how to read. Categories, respondents, and verbatims here are illustrative for this preview; every live portal is built entirely on your own data.

Part one

Your own research

Research Library · Synthesis

"What media channels drive the most brand awareness in our category?"

HIGH CONFIDENCE 8 slides · 1 project FRESHNESS · 2.8 yr
Top information-gathering channels (Media Profile Study, n=1,000)
Overall: Online research (62%), TV / streaming (55%), promotional emails (52%), and online ads (49%) led across the category.
Higher-income buyers indexed heavily on online research and YouTube; value buyers over-indexed on retailer emails, in-store displays, and print mailers.
Strategic implication
For the core buyer, online research, TV / streaming, and YouTube are the highest-reach and highest-influence channels; retailer email and in-store remain the most efficient path to value shoppers.
Sources
1 Media Profile Results PPT Slide 47
2 Media Profile Results PPT Slide 52
3 Media Profile Data Set XLSX Tab: Channels
Ask Your Data · Generated table↓ WORD / SLIDES

"Compare purchase intent for Concept A ($129) vs Concept B ($149), by segment."

HIGH CONFIDENCE n=1,817 · 1 dataset
SegmentConcept AConcept BLift
Core Enthusiast58.0%61.5%+3.5
Practical Owner41.0%47.5%+6.5
Premium Seeker34.0%36.0%+2.0
Value Shopper28.0%25.5%−2.5
Overall33.0%38.0%+5.0
Headline read: The aggregate lift is concentrated in Practical Owners. Core Enthusiasts are near their ceiling; Value Shoppers move the wrong way at the higher price.
Top-2-box = "definitely" + "probably would purchase." Within-segment rates from the concept-test dataset; export carries the full table and chart into a branded deck.
Part two

Your synthetic panel

One-on-One · Synthetic conversation
W
Wade T.
51 · Youth soccer league director · suburban Midwest
Responsible for weather calls on 14 fields · safety over gadgets
Heavy user
I want to show you a handheld lightning detector we're considering. What's your honest first reaction?
Handheld lightning detector shown to the respondent
Stimulus shown in-conversation
WADE T.
First thing I look for is the distance bands — the 0–10 and 10–25 mile readouts are what I actually use, because our rule is clear the fields at ten miles. I like that the strike count is right there; that's what I show a parent who's arguing with me. What I can't tell from a photo is how loud the alarm is on a windy field and whether the battery lasts a full Saturday of games. That's what decides it.
Fair. What would make you buy this over the phone app you're using now?
WADE T.
Honestly? Something I can clip to the fence that doesn't depend on a cell signal or my phone battery. An audible alarm I can hear from the far touchline, and a log I can hand the league board when someone asks why I called a game. I'll pay a hundred, even one-fifty, if it's the thing that keeps me from making that call by eye. Get those right and the price stops being the conversation.
Survey · Brand preference↓ CSV / PDF

"Of these four fresh dog-food brands, which is closest to your go-to?"

50 respondents · single-select · ~90 sec runtime
By brandBy reasonBy driver
#1Canine Choice fresh dog food
Canine Choice
Ingredient-led, "real food" positioning
32%n=16
#2Happy Paws fresh dog food
Happy Paws
Small-dog specialist, premium fresh
28%n=14
#3Nature's Harvest fresh dog food
Nature's Harvest
Sustainable-meat positioning
24%n=12
#4Cascade Provisions fresh dog food
Cascade Provisions
Homestyle, grain-free
16%n=8
"Why that brand?" — themed open-ends
My dog finishes the bowl — "she actually eats it"20 (40%)
"She cleans the bowl every time. With the last brand half of it went in the trash."
Ingredients I can read / vet recommended13 (26%)
"I can read every ingredient on the label, and my vet didn't wince when I showed her."
Calibrated synthetic panel. Product names and findings shown are illustrative — sample conditions, not a fielded engagement.
Focus Group · Transcript

"What do you need most from a robot vacuum?"

Moderator: Let's start with your honest, real opinions — no right answers.
Dana T.Pragmatic
It just has to not get stuck. My last one wedged under the couch every single day — if I'm rescuing it constantly, what's the point?
Russell M.Cautious
Pet hair is the whole reason I'd buy one. If it can't handle two shedding dogs, it's a gadget, not a tool.
Priya N.Warm
Quiet matters more than I expected — I work from home, and I want to run it without wearing headphones to survive it.
Emotional tone: Pragmatic skepticism · Mild frustration · Cautious optimism
Focus Group · Report↓ PDF / CSV
3 groups · 12 participants · market-representative · same study, report format
TOPLINE: Reliable navigation and pet-hair pickup were the most consistently cited needs — 8 of 12 raised getting stuck or poor mapping, 7 of 12 raised pet hair. Quiet operation emerged as a surprise must-have, cited by 5 of 12 across all three groups.
KEY DIFFERENCES
• Navigation / not getting stuck cited by 8 of 12, consistent across all groups.
• Pet-hair pickup ranked second — strongest in Groups 1 and 3.
• App control mattered most to the core segment; battery life a minor theme.
OVERALL SENTIMENT
Pragmatic 41% · Cautious 22% · Frustrated 12% · Warm 12% · Other 13%

The question isn't "can you build a synthetic panel." It's "does it answer like my customers."

Anyone can stand up a synthetic panel. Pervasive Insights™ is calibrated to your own commissioned research — and the calibration is auditable. Your human data is layered into every result: where you have asked the same question before, the synthetic result comes back cell by cell against the human answer; where you have asked the same type of question many times, it is placed on the range your real customers gave, and anything outside that range is held for review rather than reported. Below: the same questions run through human respondents and through the synthetic panel. Distributions match within ~3 percentage points on every cell.

Brand Awareness · 4 brands
Avg Δ 2.5pp
"Which of these brands have you heard of? (Select all that apply)"
Human survey · n=1,500 Synthetic panel · n=50
Brand AΔ 3pp
Human67%
Synthetic64%
Brand BΔ 3pp
Human42%
Synthetic39%
Brand CΔ 3pp
Human18%
Synthetic21%
Brand DΔ 1pp
Human9%
Synthetic8%
Human field: 14 days Synthetic: 90 seconds
Purchase Intent · 5-pt scale
Avg Δ 2.0pp
"Based on this description, how likely would you be to purchase this product?"
Human concept test · n=400 Synthetic panel · n=50
Definitely would purchaseΔ 3pp
Human8%
Synthetic11%
Probably wouldΔ 2pp
Human28%
Synthetic26%
Might or might notΔ 2pp
Human35%
Synthetic33%
Probably would notΔ 2pp
Human19%
Synthetic21%
Definitely would notΔ 1pp
Human10%
Synthetic9%
Human field: 21 days Synthetic: 90 seconds

The calibration report card

Every client sees exactly where the panel lands against their real survey data — topic area by topic area, segment by segment. Nothing is hidden.

92%
Pass rate
cells within ≤5pp of human
~3.5pp
Average error
per cell, across topic areas
19 × 5
Calibrated
topic areas × segments
Topic area Core Enthusiast Practical Owner Premium Seeker Value Shopper Newcomer
Functional perception3.23.92.43.12.9
Emotional perception3.72.83.33.72.2
Brand preference1.31.51.51.31.4
Purchase intent3.44.93.24.63.3
Consideration set5.95.84.26.25.1
Price sensitivity2.63.32.72.93.1
Green = within ≤3.5pp of the real human distribution · amber = wider, flagged for the next tracker wave. Each new wave of human research tightens the panel further. (Illustrative sample — topic-area names shown; client data drives the live card.)

Behind the report card: roughly 1,080 calibration targets across all 19 marketing topic areas, benchmarked against 40+ years of real human respondent data and re-verified with tens of thousands of independent statistical checks every calibration pass. The synthetic panel doesn't replace your human research — it lets you query that research base instantly between waves.

From your historic research to on-demand outputs.

No new fielding required to start. We build the synthetic panel from data you already have — then keep tuning it as new tracker waves come in.

Human research first, machine learning built on top: the Pervasive Insights™ loop
Human research first. Machine learning built on top. Every loop makes the next study count for more.
1

Your research

We index every commissioned study you've run — surveys, focus groups, trackers, ad tests. Years of source data, all in one searchable place.

2

Calibration

We build identity profiles calibrated against your actual respondents. Every dimension on every profile is anchored in real data, not guessed — and the error is reported.

3

Synthetic panel

6,000 calibrated identities sit in your private portal, ready for any question. They speak in your customers' voice — drawn from your real verbatims.

4

On-demand research

Search the library, query your data, or run an interview, survey, or focus group in 90 seconds — the same deliverable you commission today, in minutes.

What comes out. Any conversation becomes a Word one-pager or analyst memo, or goes straight to slides. Library search downloads only the referenced slides. Synthetic surveys download the dataset and the question-by-question summaries. Synthetic focus groups download the full transcript and a summary report. Everything you export is kept on a Deliverables shelf in the portal, by date.

Calibrated. Auditable. Cited back to source.

Every dimension on every identity profile is anchored in real survey data. Calibration error is reported per topic area, per segment. Nothing is generated from thin air.

~3.5pp
Average distribution error against real survey data, across all 19 marketing topic areas.
6,000
Calibrated identity profiles per client, drawn to match the actual buyer population.
40+ years
Of real human respondent experience across thousands of research engagements, behind every calibration.
~90 sec
From question to readout. Run as much research as your team can absorb, between waves and before them.

The portal is one part of a research firm's toolkit.

Pervasive Insights™ is the AI suite of City Research Solutions, a consumer research firm that has fielded surveys and focus groups since 1979. The library and the calibrated panel you see here sit alongside a third tool, and all three draw on the same material: research you have fielded with real people.

Your research, working
Research Stack: Library search & Data synthesis. Every report deck searchable in plain English and cited to the slide; raw datasets questioned by segment, region or wave. Details →
Your customers, simulated
Synthetic customers: surveys, focus groups, one-on-ones. 50 to 500 calibrated synthetic panelists drawn from your own customers, with a Calibration Report Card for every marketing topic area. This site. Details →
Your market, watched
Competitive Intelligence. Monthly competitive sweeps, social listening and grey-market watch across the retailers and markets you compete in. Details →
Human research, first
City Research Solutions. In-facility, online, real-time, retail and quantitative research, in person and online, since 1979. Visit CRS →
Need a real focus group, survey or shop-along? The firm behind the portal runs them: In-Facility Online Real-Time Retail Quantitative
City Research Solutions facility

Forty-plus years of commissioned consumer research, made queryable on demand.

Pervasive Insights™ is a product of City Research Solutions — a Midwest-based consumer research firm with more than four decades of category experience.

CRS has fielded thousands of commissioned research engagements across categories spanning workwear, outdoor power equipment, outdoor furniture, personal care, food and CPG, tools and hardware, retail, and more. Every Pervasive Insights™ client portal is built directly on top of that client's own commissioned research history — and calibrated against 40+ years of real human respondent experience.

The synthetic panel is not generic. It's yours. And it is the middle of a three-part suite: your research library on one side, a monthly read on your competitive market on the other, with the firm's human research services underneath all of it.

A City Research Solutions product · Pervasive Insights™ 2026 · Visit CRS →

Schedule a 30-minute walkthrough.

Tell us a little about your category and the kind of research questions you wrestle with most often. We'll send back available times within one business day.

Questions we get asked:What is Pervasive Insights™?How much existing research do you need?How is a synthetic panel calibrated?Synthetic vs. traditional panelsHow accurate are synthetic respondents?Synthetic research for CPG brandsMethodology Q&A